J. LeConte — Phenomena of Binocular Vision. 103 



seen at A and C, and the phantom plane would be P' P'. Such 

 it would be by geometric construction and thus it is always 

 represented. Such would indeed be the case of the retinas 

 were a 'plane parallel with PP. But the retinas are spherical 

 concaves, and for this reason when bb fall on corresponding 

 points, viz: the central spots bb, a a and cc do not fall on ex- 

 actly corresponding points ; for it is evident on inspection 

 that the retinal points a a and cc are nearer together thanbb. 

 Therefore they form homonymously double images and are 

 therefore referred to, and combined at, a point farther away than 

 B. Therefore the phantom surface must appear curved from 

 side to side, as shown in the dotted line. 



Prof. LeConte Stevens, in an article published in 1882,* ex- 

 plains this phenomenon of transverse curvature of the phan- 

 tom image by "changes in muscular tension," or as I would 

 say, changes in axial adjustment. But (1) no such changes 

 are necessary to perceive the phenomenon. The curvature is 

 best seen with the point of sight fixed and is doubtless due to 

 the slight (perhaps not consciously perceptible) homonymous 

 doubling of the images of points right and left of the point of 

 sight. And (2) in any case the fundamental explanation is 

 found only in the law of corresponding points; for this law 

 alone explains the necessity of the changes in axial adjustment, 

 which are necessary to combine the double images. The true 

 explanation is indeed involved in Prof. Stevens's figure (3) and 

 especially in his statement that it is the result of the concavity 

 of the retina; but he apparently does not see its necessary con- 

 nection with the law of corresponding points, and even seems 

 to doubt the validity of that law.f 



Experiment 4. — Prof. Stevens:]; has devised a beautiful and 

 ingenious method of bringing out strongly the transverse cur- 

 vature of phantom images. But the curvature brought out by 

 his method has largely a different cause from that already 

 described, as we now proceed to show. 



We have seen that the curvature in the preceding experi- 

 ments is due to the extreme, but varying, obliquity of the parts 

 combined. Now it is evident that this obliquity may be more 

 easily gotten and the combination effected without extreme and 

 straining convergence, by simply bending the plane at the 

 middle line and inclining the two halves in opposite directions. 

 This method has the additional advantage of allowing inclina- 

 tions backward or forward and thus producing reverse effects; 

 and also of allowing combination of the oblique surfaces 

 beyond the plane of the object, either by naked eyes or by 



* This Journal, vol. xxiii. p. 297 and seq. f Ibid., pp. 300 and 301. 



% Ibid., p. 298. 



